1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

Converter experiment

Discussion in 'Recording' started by BobRogers, Nov 3, 2007.

  1. BobRogers

    BobRogers Well-Known Member

    So I did a little experiment today that I've been meaning to do for a while. I've been recording my bass tracks through the SPDIF outs of my Line6 Bass POD xt-pro to my digi 002r. Today I decided to record the analog out simultaneously. I'm not completely familiar with the block diagrams of either machine, but I assume that the biggest (if not the only) difference between the two tracks should be that the digital direct eliminates two conversions.

    So to hear the difference I inverted one signal and nudged the analog forward to correct for the slight latency. This lets me hear the actual difference (which is pretty significant) and balance the volumes by finding the spot where the difference has lowest volume. As you would expect the difference is all in the high end. The solo analog signal is not at all unpleasant - the high frequency roll off sounds pretty nice in this particular riff (I was working on the intro to So What). But the difference is definitely audible - though I'm always a little suspicious of non-blind tests like this.

    Can anyone suggest (inexpensive) ways that I could improve this experiment? Things I should be listening for in order to understand digital conversion artifacts?
  2. Cucco

    Cucco Distinguished Member

    So you took the analog signal from the POD and the digital signal at the same time into the 002r?

    Wouldn't the analog signal have gone through the POD's AD and the DA conversion first and then through the AD conversion in the Digi?

    Of course, I don't think this one extra step (err...2) made a huge difference, but a difference nonetheless.

    Are you tied to doing your conversion tests using bass?

    When I'm comparing A/D converters, I'll often use a normalled patchbay to send the same signal to both converters and then send the input into my DAW and reverse the phase. Of course, there's no way to tell which one is leaving the artifacts. (Well there is...but you have to bounce the remaining artifacts from the summed and reversed signal and then experiment by adding it back to each raw signal - phase reversed, bounce it down and then see which one causes a total cancellation....too confusing?)

    I can't say I've ever done this for anything more than my own personal edification. It's interesting to hear just what is left over. What's really interesting is when you find that "inexpensive" converter that leaves hardly any artifacts when compared with one of those really expensive converters....
  3. BobRogers

    BobRogers Well-Known Member

    Right. I realize I can't isolate either converter this way, I have to listen to the effect of both. But, frankly it's the first time I've done a test that isolated converters from the rest of the system. I'm not tied to using bass for the test, but it is a Bass POD and I was recording bass tracks this morning. I was planning to add some electronic keyboard tracks, and I could try putting them through the POD (even thought the bass POD is mono).
  4. BobRogers

    BobRogers Well-Known Member

    Yes, too confusing. Run that by me again, please.
  5. Cucco

    Cucco Distinguished Member

    Sorry...let me explain further.

    Take the original 2 comparisons (wave 1 and wave 2). Flip the phase on one of them.

    You're left with the remnants of the comparison between the two (some HF stuff and maybe a single note here and there elsewhere in the spectrum)

    Take wave 1 that you had and add the remnants to it. Flip the phase of the remnants (you won't hear much of a change at all - if any). Now, bounce that track down to...let's call it Track A now.

    Now, load up your daw and put Track A down with Track 2. Flip the phase on one of them. If there is no sound at all, then you learned that Track 1 was the track that contained "extra" or errant information. If you hear sound, then you know that it *could* have been Track 2.

    The likely scenario is, however, that both tracks contained information that wasn't in the track. However, if one AD converter appears (audibly) to contain additional HF information, this AD converter may be considered to be "brighter" or more sterile (depending upon your verbage).

    The opposite could be said for the other AD converter.

    These are, of course, gross generalizations but nonetheless it's what I suspect you're looking for between the comparitors.

    On a different note...I'll be in the Charlottesville area Tuesday afternoon evening and again later in the month a few times recording UVA. You're welcome to come sit in on the sessions and if your bari-sax tooting daughter is interested too, I think I could make some room.

    Of course, it's hard to bring a Hokie into Cavalier territory without starting a war....

    PS - it's their marching band that I'm recording.

  6. BobRogers

    BobRogers Well-Known Member


    Tuesday will be bad for us (Alice will be down in Georgia auditioning for their music school and I'm doing senior exit interviews) but when are you recording later in the month. I will not wear any Hokie regalia, especially since we are currently tied with the Hoos for the lead in the ACC coastal division. We play them in C'ville in three weeks.

    If I understand your experiment though, I would expect your mixed track to cancel track 2. Here is how I understand it:

    track 1 = base signal + noise 1
    track 2 = base signal + noise 2

    remnants = track 1 - track 2 = noise 1 - noise 2

    track 1 - remnants = base signal + noise 2 = track 2

    Unless I'm missing something I don't see how it tells me anything about the relative sizes of noise 1 and noise 2.
  7. Cucco

    Cucco Distinguished Member

    Cool. I'll let you know the next time I'm there.

    I think I'm confused now about my experiment....

    Let me see if I can put some graphics to it to help out.

    I think the above explanation though is over complicating it - adding noise where there shouldn't be.
  8. BobRogers

    BobRogers Well-Known Member

    Maybe my language is confusing. The way I am thinking about it, for a given analog signal there is a theoretically idea digitization - one that produces a band limited version of the analog signal exactly. I am calling any deviation from that "noise." Probably not the best term to use, maybe "conversion error"?

Share This Page